July, 1963 289

THE FOSSIL OF

By ALDEN H. MILLER

When the abundant fossils collected in recent in the Lake Eyre basin of Australia were first tentatively assorted, the most surprising discovery was that of flamingos (Stirton, Tedford, and Miller, 1961:35). This type of bird is absent from the continent of Australia today and flamingos do not occur at points nearer than north- west ; there are records in historic times from Ceylon, and their bones occur along with remains of the Dodo in Mauritius. The further study of our collections, especially those obtained in 1961, shows that flamingos were present not only in mid-Tertiary time as previously noted but also in the Pleistocene, although strangely they did not appear among the Pleistocene collections DeVis ( 1906) reported on from Cooper Creek in the Lake Eyre area. In all, there are four represented, three of which are here described as new. In two of the faunas two kinds of occur together. This is not an unusual situation in this order of . Three specieslive today in the high- land playa lakes of the arid sections of , and the presence of several species to- gether in the of was well documented by Milne-Edwards (1867-1871). Flamingos in general are well represented in the fossil record, both the typical group, the Phoenicopteridae, and the extinct Paloelodidae, a group of birds that is straight-billed and also less slim and elongate than the true flamingos. The environ- ment in which such birds live, namely relatively shallow, muddy lakes of wide expanse, has been particularly favorable to the preservation of their bones. Thus the record of the order extends back to the and geographically to many parts of and North America. Fossils and modern occurrences are not known from Southeast Asia or the’East Indies. The lack of fossil records for South America and Africa prob- ably reflects in part the.relatively slight attention given to the fossil birds of those continents. The support of the National Science Foundation in 1961 and 1962 made possible further field work in Australia and the processing of collections, both of the essential Recent bird skeletons and of fossils. The study of fossil vertebrates in South Australia has continued to receive the generous support of the South Australian Museum and its staff. We were especially aided in 1961 by Norman B. Tindale, Paul F. Lawson, and Harry J. Bowshall. R. A. Stirton, Richard H. Tedford, and Virginia D. Miller of our field party provided critical assistance during the 1961 expedition. Stirton particularly has given constant support and encouragement to the investigation of the Australian fossil birds and has supplied details concerning the collecting localities. I am indebted also to Philip S. Humphrey of the United States National Museum for the loan of skeletons of Phoenicopterus ruber.

DESCRIPTION OF MATERIAL

Phoenicopterus novaehollandiae, new species

Type.-Right tarsometatarsus, the distal end essentially complete, although with many frac- tures proximal to the articular surfaces; about 95 per cent of shaft represented; no. P13648, South Australian Mus.; locality no. V6150, Univ. Calif. Mus. Paleo., Lake Pitikanta; Ngapakaldi Fauna, middle Tertiary, probably late Oligocene or early Miocene; figures la, c, 2a. Type locality.-Etadunna Formation in escarpment on west side Lake Pitikanta about 600 yards from south end; Lake Eyre region, South Australia; green to greenish-brown mudstone below upper gray calcareous mudstone and near contact with lower gray calcareous mudstone. 290 THE CONDOR Vol. 65

Diagnosis.-Similar in foot structure to Phoenicopterus ruber (includes Phoenicopterus ruber rosew) but scar for articulation of metatarsal I present; distal tarsal foramen apparently not per- forating plantar surface and on anterior surface situated more proximally; trochlea IV closer to trochlea III; posterior extension of trochlea II narrower mediolaterally and proximodistally. Size within limits of variation of ruber but near maximum of range.

Analysis and comparison.-The species novaehollandiae has the same general con- figuration, slender construction and articular system of the foot as that of the modern , P. ruber, of Africa, southern Europe, and the West Indies. The most interesting departure is in the presence of the articulation for a first toe which can only mean that this digit was of a size and functional importance much greater than in modern flamingos. In the present-day Phoenicopterus it is reduced to a length of about 18 mm. and seems to have little use. In the Phoeniconaias it is further reduced and in Phoenicopawus it is lacking. Even in the first two, there is no flattened scar for articulation at the point of attachment on the medial ridge of the plantar surface, the toe being held at that point rather loosely by weak ligaments.

:I j

a b C d

Fig. 1. Distal ends of right tarsometatarsi of flamingos, natural size. a, type of Phoenicopterw novaehollandiae, medial view; b, modern Phoenicopterus ruber roseus, no. 289137, U. S. Nat. Mus., medial view; c, type of P. novaehollandiae, plantar view ; d, P. Y. yoseus, plantar view. Drawings in figures 1 to 6 by Augusta Lucas.

The other features of Phoenicopterus novaehollandiae in respect to the trochleae reflect a lesser spread of the toes and perhaps weaker support for the medial bracing ligaments of the inner toe. These differences are of small magnitude and probably lim- ited functional significance but are consistent departures from the configuration in the July, 1963 FOSSIL FLAMINGOS OF AUSTRALIA modern ruber in that none of the considerable sam- ple of rubier shows them. They are therefore useful speciescharacters. Phoenicopterus novaehollandiae had an overall length and slenderness equivalent to specimen no. 289737, U. S. Nat. Mus., a large male P. ruber (figs. lb, d, 2b). This length is judged from the dis- tance from the distal end to the beginning of the expansion of the proximal end, which expansion is present in the fossil even though the proximal artic- ular surfaces and the hypotarsus are entirely lack- ing. The shaft, which was lying in place, but in a much fragmented condition, has been fitted together, all junctions but one being natural contacts. Because of the one gap in mid-shaft, which is believed to be small, the estimate of length is minimal, but it is not likely to be more than 5 per cent below the true value. Five extinct species of flamingos of the genus Phoenicopterus have already been described from other parts of the world. Two of these are very small species,neither of which is represented by the distal end of the tarsometatarsus. By reason of small size at least, they are not confusable with novaehollan- diae. These two are Phoenicopterus minutus Howard (19.55) from the late Pleistocene of California and Phoenicopterus stock L. Miller (1944) from the Pliocene of Chihuahua. Phoenicopterus copei Shufeldt ( 1892 ) of the Ple- istocene of Oregon, consisting of several unassociated parts, does include a distal segment of a tarsometa- tarsus as designated type material. Howard (1946: 157-158) did not find characteristics in this part of the tarsus either in size or shape which separated copei from ruber, in the inclusive sense, although there were features of the tibiotarsus that did so dis- tinguish it. The figure of the tarsometatarsus pre- pared by Shufeldt himself shows this bone in an- terior aspect to have none of the compression of the trochleae of novaehollandiae, and a completely per- forate distal foramen is present and situated as in ruber. b Phoenicopterus floridanus Brodkorb (1953) from the Pliocene of is based on a tibiotarsus but there are referred specimens of the tarsometatarsus. Brodkorb’s figure of this element shows none of the characteristic shape of trochlea II of novaehollan- Fig. 2. Anterior views of tarsometa- tarsi, X %. a, type of Phoenicop- diae. Moreover the shaft is much deeper in floridanus terus novaehollandiae ; b. modern and the distal foramen is perforated and situated Phoenicopterus ruber roseus, no. more as in ruber. 289737, U.S. Nat. Mus. 292 THE CONDOR Vol. 65

The only old world fossil of the genus Phoenicoptems which has been given a specific name is Phoenicopterus croizeti Gervais of the Aquitanian (Oligocene) of France. This was founded on a tibiotarsus. Much additional material has been referred to it, including several tarsometatarsi. Milne-Edwards (1867-1871 :pl. 81) in figuring these tarsi shows none of the features that would suggestnovaehollandiae. For example trochlea II is not compressedor tapered posteriorly and the lateral rim of trochlea IV is not produced and therefore does not reflect compression; in both these respects it is like ruber. The distal foramen clearly is placed rather far distally, not proximally as in novaehollandiae. Unfortunately the plantar surface is not figured or commented on so that evidence for the development of digit I cannot be assessed.It would be most interesting to know if croizeti as a species of similar age to novaehollandiae gave indi- cation of having a similar large hind toe unlike the later flamingos.

TABLE 1

MEASUREMENTS IN MILLIMETERS OF TARSOMETATARSI OF SPECIES OF Phoenicopterus

Total Le$$;;:h Le$&w;h W;;Ah.agss length

P. novoehollundioe 3542 6.9 7.0 17.6 ’ P. ruber roseus, no. 289737 ( $ ) U.S.N.M.’ 356 6.2 6.9 21.6 P. ruber roseus, no. 224858 U5N.M.l 364 5.6 6.7 20.6 P. ruber roseus, no. 79026 M.V.Z. 275 5.8 6.4 18.5 P. ruber ruber, no. 140923 M.V.Z. (Galapagos Islands) 253 5.7 6.2 18.0 P. chilensis, no. 125159 ($ ) M.V.Z. 276 5.6 5.3 19.3 P. chiknsis, no. 125158 ( 0) M.V.Z. 236 5.3 4.8 16.7

1 Largest two skeletonsof 16 of this form in U. S. Nat. Mus. * Estimate extrapolated from comparisonwith no. 289737 U. S. Nat. Mus. 3 Dimension may be somewhat reduced (less than 10 per cent) by distortion in prexrvation.

Phoeniconotius eyrensis,new genus and species

Type.-Distal end of left tarsometatarsus, consisting of trochleae II and III and adjoining foot area and fragmented trochlea IV, the shaft lacking; proximal phalanx of left digit III and proximal phalanx of right digit IV; no. P13649, South Australian Mus.; locality no. V5763, Univ. Calif. Mus. Paleo., Lake Palankarinna, Ngapakaldi Fauna, middle Tertiary, probably late Oligocene or early Miocene; figure 3. Type locality.-Base of escarpment on west shore of Lake Palankarinna between turtle quarry (V.5762) and Perikoala locality (V5375) ; Lake Eyre region, South Australia; float from Etadunna Formation with other bird, reptile and fish remains derived from green mudstone and fine-grained sandstone units, about 10 feet in thickness, above lower limestone member. Diagnosis.-Resembles Phoenicopteius but articular surface of trochlea III on plantar surface truncated proximally, not long and tapered; trochlea II with smooth, broad and conspicuous con- cavity bordering articular surface proximally, the medial ridge delimiting it sharp and displaced far toward medial surface; prominent scar present for articulation of metatarsal I; trochlea III more robust and wider; basal phalanx of middle toe more robust but relatively shorter, the plantar fossa at its base deeper. Size somewhat larger than largest males of Phoenicopterus ruber, at least in breadth and depth of foot. Association.-The material representing Phoeniconotius was picked up at one spot on the surface of the formation but was not in articulated position. The base of the middle toe fits the articular surface of trochlea III and belongs to the same side of the body. The basal part of toe IV appears from its shape to pertain to the opposite foot and because of this and since trochlea IV is incomplete, its relation to the other parts July, 1963 FOSSIL FLAMINGOS OF AUSTRALIA 293

Fig. 3. Distal end of left tarsometatarsus and base of toes of Phoeniconotius eyrensis, type, natural size. cz, medial view; b, plantar view; c, anterior view. The outline of the basal phalanx of digit IV is derived by reversing the drawing of the example of the right phalanx. cannot be clearly demonstrated. All three units are of the same gray color and are highly mineralized. Analysis and comparison.-Phoeniconotius like other phoenicopterids has the characteristically elevated trochlea II which is deflected strongly toward the plantar surface. GYW has some of these features, but trochlea II of this genus is much more elevated and the posterior spur of trochlea II is narrow and set off sharply from the articular surface. In the family Paloelodidae trochlea II is of very different shape in medial profile, being broad and rounded and having no true posterior spur as in Phoenicopterus and cranes. Phoeniconotius in none of these respectsapproaches Paloelodus and Megapalo- elodus and is instead a typical flamingo of the Family Phoenicopteridae. However, in general massiveness,as judged from the middle trochlea, it is the equivalent of Mega- paloelodus connectens from the Miocene of South Dakota (A. H. Miller, 1944: 86, fig. 1). The structural features of Phoeniconotius suggestthat it was less equipped to swim than Phoenicopterus and its small modern relatives. This is shown by the articular surface of the plantar aspect which is less extended to accommodate extreme flexion of the toes. In swimming groups generally this surface is well extended whereas in more terrestrial or walkers, such as cranes, it is not. Phoeniconotius may then have been a flamingo less adapted than the other genera for deep water wading or swim- ming. Its relatively short, stout basal phalanx of digit III further suggeststhis. As in Phoenicopterus novaehollandiae, the first toe in Phoeniconotius was evidently better developed than in modern flamingos, for the articulation for metatarsal I is con- 294 THE CONDOR Vol. 65

TABLE 2

MEASUREMENTSIN MILLIMETERS OF PARTS OF FEET OF GENERA OF FLAMINGOS Tarsometatarsus: Phoeniconotiw Phoenicom.rus n&r Mc~o~oloelodus (no. 224858U.S.N.M.) Greatest anteroposterior dimension of trochlea II 15.0 13.9 16.0 Greatest anteroposterior dimension of trochlea III 12.5 11.1 13.7 Greatest width of trochlea III 10.5 8.3 9.0 Distance from tip of trochlea III to plantar end of articular surface 11.1 12.3 12.8 First phalanx digit III: Length 45.5 48.8 Proximal width 9.9 9.1 Proximal depth 11.5 10.8 Distal width 6.3 6.2 Distal depth 6.4 6.2 First phalanx digit IV: Length 31.9 30.8 Proximal width 8.5 7.3 Proximal depth 11.0 9.9 Distal width 4.7 5.3 Distal depth 6.7 7.0 spicuous and broad. Milne-Edwards’ figures (op. cit.: pl. 82; pl. 87, 2) suggest this toe was well developed in Paloelodus, but the evidence for this is not available for the Oligocene Phoenicopterus croizeti, as already noted. A fairly large hind toe and articu- lation are present also in the cranes and of course in the and . In summary, then, Phoenicolzotiuswas a large, massive, true flamingo, less special- ized for swimming than other flamingos and possibly more adapted than others for shallow-water and terrestrial locomotion. Without knowledge of its linear dimensions, we cannot judge whether its legs were elongate in such great degree as in other members of the Phoenicopteridae. The details of its foot’ and toe articulation suggestno signifi- cant approach to or link phyletically with the massive, relatively shorter-legged fla- mingos of the family Paloelodidae.

Phoeniconaias gracilis, new species Type.-Distal end of left tarsometatarsus, complete except for some abrasion of anterior surfaces of trochleae II and III; plantar surfaces and mcst of articu!ar surfaces well preserved; no. 13650, South Australian Mus.; locality no. V5772, Univ. Calif. Mus. Paleo., Lake Kanunka, Kanunka Fauna, early Pleistocene; figures 4a, c. Type locality.-Katipiri Sands at northwest corner of Lake Kanunka; Lake Eyre region, South Australia; coarse to medium white and ferruginous stained quartz, stream-channel sands with clay balls, pebbles, coprolites and abraded bones; channel sands have cut through Tirari Formation and rest on green mudstones of Etadunna Formation (Stirton, Tedford, and Miller, 1961). Diagnosis.-Smaller and more slender than Phoeniconaias minor; posterior extension of trochlea II much less rounded and of lesser proximodistal dimension; articular surface of trochlea IV less extensive proximally on plantar surface. Analysis and comparison.-Among living flamingos, the small speciesare separated generically from Phoenicopterus as and Phoeniconaias on the basis of substantial differences in the rostrum and skull. Generic differentiation in the tarso- metatarsus is slight, but the small species have a more tapered and relatively longer articular surface at the plantar base of trochlea III. This is particularly evident in Phoeniconaias minor although I have been able to compare only one of the two species July, 1963 FOSSIL FLAMINGOS OF AUSTRALIA 295

a

Fig. 4. Distal ends of left tarsometatarsi of flamingos of the genus Phoeniconuias,natural sire. a, type of Phoeniconaiusgracilis, medial view; b, modern Phoeniconaiasminor, no. 133411, Mus. Verb Zoo]., medial view; c, type of P. grucilis,plantar view ; d, P. minor, plantar view. of Phoenicopurrus, namely and&s. Furthermore, in Phoeniconaias trochlea II is rela- tively shallow and less rounded in the four specimens examined than in Pkoenicoparrus and&us, Phoenkopterus ruber, and Phoenicopterus chilensis. In these respects the fossil gracilis accords with Phoeniconaias minor or exceeds it in differentiation from the other genera. These circumstances and the greater geographic proximity of Phoeni- coniziusof Africa and northern India favor placing gracklis in this genus even though

TABLE 3

MEASUREMENTS IN MILLIMETERS OF THE TARSOMETATARSI OF SPECIES OF Phoeniconaias

P. gracilis P. minol type p, no. 0, no. 0, no. 8, no. 133408 133409 133410 133411 MVZ MVZ MVZ MVZ Greatest width across trochleae 13.o 14.7 15.3 15.8 14.8 Width of trochlea IV 3.3 3.9 4.0 4.4 4.1 Greatest anteroposterior dimension of trochlea III 8.8 8.4 8.7 8.7 8.5 Greatest anteroposterior dimension of trochlea II 8.5 9.1 9.2 9.6 9.2 Minimum width of shaft 4.7 5.6 5.5 6.0 5.5 Maximum depth of shaft 5.6 5.7 5.3 5.8 5.6 Total length _... 205 214 219 230

one would prefer, were it possible, to make the allocation on the basis of the sub- stantial skull differences known in the modern types. Other small fossil flamingos already mentioned (p. 291) are not known in respect to the foot area of the tarsometatarsus and occurred in western North America. Both appear to have been significantly smaller than females of Phoeniconaias minor but not quite as small as gradis which is about 10 per cent smaller than the known minimum of minor. It is safe to say that gracilis was as small as any flamingo thus far recorded and it has features of configuration in the foot distinct from those of other flamingos in which that part is represented. Additional material.-A fragmentary right tarsometatarsus consisting chiefly of trochleae II and III, no. 60561, Univ. Calif. Mus. Paleo., was found at the same local- 296 THE CONDOR Vol. 65

, ity as the type in a subsequent . It shows the same features of trochlea II, the slight departures from the type apparently being due to some abra- a b sion. It was of the same small size and slender build. Referred materiaLThe distal end of a right tibiotarsus (fig. Sb), no. 56887, Univ. Calif. Mus. Paleo., from the same locality, differs from Phoe- niconaias minor in smaller size in the same degree as does the type tarsometatarsus. It matches minor well in configuration of the anterior sur- face, which area, including the tendinal canal, is Fig. 5. Distal ends of tibiotarsi of fla- well preserved. The condyles are abraded exten- mingos of the genus Phoeniconaias, sively on the posterior surface. Because of corre- anterior view, natural size. a, Phoeni- conaias minor, no. 133411, Mus. Vert. spondence in size to gracilis, it is referred to that Zool.; 6, fossil no. 56887, Univ. Calif. species. Mus. Paleo. referred to Phoeniconaias The distal end of a left humerus, no. 56882, gracilis. Univ. Calif. Mus. Paleo., from the same locality is also abraded, especially on the posterior surface, but shows enough of its configuration to place it as a flamingo. Its size is the same as that of specimen no. 133411, Mus. Vert. Zool., of Phoeniconaias minor. It is therefore not as extremely small as gracilis as known from the leg bones. However, it possibly falls within the limits of variation in size of that specieswhich of course might not have shown in the wing the very slim build and small size that it did in the leg. It seemsbest, therefore, to refer the fragmentary humerus to gracilis.

Phoenicopterus ruber Linnaeus

Material.-Fragmentary distal end of a right tarsometatarsus lacking part of trochlea IV and the posterior spur of trochlea II; no. 60583, Univ. Calif. Mus. Paleo.; locality no. V5772, Univ. Calif. Mus. Paleo.; in situ in Katipiri Sands, Lake Kanunka, Kanunka Fauna, early Pleistocene; figure 6a. Distal end of left humerus, well preserved but part of entepicondyle abraded; no. 56360, Univ. Calif. Mus. Paleo.; locality no. V5866, Univ. Calif. Mus. Paleo.; Cooper Creek, Malkuni Fauna, late Pleistocene ; figures 6b, c. ,I- I I ’ I

a b C

Fig. 6. Pleistocene fossils of Phoenicopterus vuber, natural size. a, distal end of right tarsometatarsus, no. 60583, Univ. Calif. Mus. Paleo., Kanunka Fauna, plantar view; b, distal end of left humerus, no. 56360, Univ. Calif. Mus. Paleo., Malkuni Fauna, anconeal view ; c, same as b, palmar view. July, 1963 FOSSIL FLAMINGOS OF AUSTRALIA 297

Localities.-For locality V5772, see p. 294. Locality V5866: site 14, Katipiri Formation on north bank Cooper Creek across bend of creek west of Malkuni waterhole and Emu Camp (Stirton, Ted- ford, and Miller, 1961) ; highly ferruginous and cross-bedded sands with interbedded gray, yellow-brown, dark-red, and green clay bases; overlain by 20 feet of dune deposits; base of forma- tion not observable.

Characters.-The tarsometatarsus shows the diagnostic shape of trochlea II of Phoenicopterus in respect to the articular surface and ligamental scars, and the length of this trochlea and its plantar deflection in all respects match Phoenicopterus ruber; only the tip of the posterior spur is lacking to prevent a verification of this part. Trochlea III likewise matches ruber in all details of shape including the plantar exten- sion of the articular surface. The distal. tarsal foramen is fully perforated and of size and position normal for ruber.’ In size the fossil is a close match for P. ruber no. 224858, U.S. Nat. Mus. (see table 2, p. 294). The parallel measurements of no. 60583 are: greatest anteroposterior dimension of trochlea III, 10.9 mm.; greatest width of trochlea III, 8.2 mm. The humerus, no. 56360, shows configurations and ligamental scars of the lateral surface of the ectepicondyle identical with those of Phoenicupterus ruber. On the an- coneal surface, the external tricipetal groove is broad, deep and extended far proximally as normal in flamingos. The palmar surface shows the characteristic deep brachial de- pression with an elevated and elongate scar for the anterior articular ligament. The size of this fragment accords well with 140923, Mus. Vert. Zool., a male Phoenicopterus ruber ruber, smaller than the large males of the old world race, but well within the range of variation of that race. The fossil is much too large for Phoeniconaius. Thus both the early Pleistocene and late Pleistocene representatives of a large flamingo show nothing to suggest any departure in osteologic detail or size from the modern Phoenicopterus rub.ey, and they may properly be assigned to that species.

DISCUSSION A question at once posed by the record of fossil flamingos in Australia is why these species became extinct. Explanations of extinction are never more than speculative for any group of , but some possibilities may be suggestedfor these flamingos. The Tertiary species, Phoeniconotius eyrensis and Phoenicopterus novaehollandiae, were both very specialized wading types that must have functioned much like modern fla- mingos. Indeed the latter especially was so similar in structural design as to suggest that it had very similar methods of foraging. If anything it was more of a stilt-like wading type than the modern specieseven though it had a larger hind toe. In the early Pleistocene, the two speciespresent were in one instance identical with a modern form and in the other a very slender, miniature counterpart of the living small flamingos. They, even more surely than the Tertiary flamingos, must have had the same ecologic requirements as the modern species. It should be noted further that the number of remains in the early Pleistocene of South Australia is considerably greater than in the much larger general collection of bird bones from the late Pleistocene in which we have found but a single flamingo bone. These several circumstances point to the presence of good shallow water lakes in interior Australia up through early Pleistocene in which flamingos could wade, forage, and presumably . Probably in order to sustain colonies, permanence of such lakes was important, if not a particular lake or shore line, a series of lakes in the desert center or its bordering semi-arid regions among which these presumably social, colonial nesters could move and always find favorable conditions. There appear to be shallow 298 , THE CONDOR Vol. 65

lakes today, such as Goyder’s Lagoon and the lake at Menindee, that might support flamingo colonies, but these may dry out too much in some years and alternate sites may be too few to sustain a colony of these large birds. Thus with the increasing aridity in the late Pleistocene the lake habitat may have become reduced in amount and de- pendability to a point below the requirement levels of these large, colonial water birds and extinction ensued. The fossil deposits from which the flamingos are derived are claystones and mud- stones along the borders of old lakes. In the early Pleistocene the matrix was a stream channel sand but this may have well been in a flat stream-mouth area bordering a lake and it was certainly not far from the former shoreline of Lake Eyre. At the time the Ngapakaldi Fauna flourished in this region, the shoreline of a much larger Lake Eyre was near the several fossil-bearing localities and some 30 miles east of the shore of the present and usually dry Lake Eyre. The of flamingos to their present stage of specialization had largely taken place by Oligocene time; the record in Europe shows some subsequent increase in bill flexure and specialization but no significant change in the feet and legs. The Australian Tertiary fossils reaffirm this attainment in the Tertiary of the modern foot structure. In the retention of a fairly large hind toe in the genus Phoenicopterus at that time we may have persistence of an earlier ancestral condition; subsequent reduction of this toe, which is of little structural significance, has taken place. Complete loss of it has been accomplished only in the modern Phoenkoparrus. Phomiconotius eyrensis was a more massively built phoenicopterid than other mem- bers of the family, paralleling in this respect the Paloelodidae, but not in any way connecting with that group. Phoeniconotius was possibly less of a deep-water or deep- mud than the other phoenicopterids, with concomitant reduction of swimming ability. As such it could represent a separate phyletic branch in that group.

SUMMARY Fossil flamingos have been found in the Lake Eyre basin of central Australia in middle Tertiary, early Pleistocene, and late Pleistocene time. The Tertiary representa- tives consist of two species, here described as new, one of which belongs to a new genus, Phoeniconotius. In the Pleistocene the modern Phoenicopterus ruber occurred in both early and late faunas and a very small, slender-footed species of the modern genus Phoeniconuias is described as new from the earlier fauna. The Tertiary speciesshowed a level of evolution of the flamingo foot equivalent to that of the modern bird but with the persistence of a larger hind toe than in the living representatives. The genus Phoeniconotius of the Tertiary was a more massive, some- what less aquatically adapted member of the Phoenicopteridae probably representing a separate phyletic branch in that family. The extinction of flamingos in Australia at the end of the Pleistocene could have been caused by increasing aridity which reduced the number and stability of large, shallow lakes.

LITERATURE CITED Brodkorb, P. 1953. A Pliocene flamingo from Florida. Nat. Hist. Misc., Chicago Acad. Sci., no. 124:1-4.

De Vis, C. W. 1906. A contribution to the knowledge of the extinct avifauna of Australia. Ann. Queensland Mus. no. 6:1-25, pls. l-9. July, 1963 FOSSIL FLAMINGQS OF AUSTRALIA 299

Howard, H. 1946. A review of the Pleistocene birds of Fossil Lake, Oregon. Carnegie Inst. Washington, publ. 551:141-195. 1955. Fossil birds from Manix Lake, California. Geol. Surv. Prof. Paper 264-J: 199-206, pl. 50.

Miller, A. H. 1944. An avifauna from the Lower Miocene of South Dakota. Univ. Calif. Publ. Bull. Dept. Geol. Sci., 27:85-100. Miller, L. 1944. A Pliocene flamingo from . Wilson Bull., 56:77-82.

Milne-Edwards, A. 1867-1871. Recherches anatomiques et paleontologiques pour servir a l’histoire des oiseaux fossiles de la France (G. Masson, Paris), vol. 2 ; atlas, vol. 1. Shufeldt, R. W. 1892. A study of the fossil avifauna of the Equusbeds of the Oregon desert. Jour. Acad. Nat. Sci., Phila., 9:389-425.

Stirton, R. A., Tedford, R. H., and Miller, A. H. 1961. Cenozoic stratigraphy and vertebrate paleontology of the Tirari Desert, South Australia. Rec. South Australian Mus., 14: 19-61.

Museum of Paleontology, University of California, Berkeley, California. Decem- ber 24, 1962. ,